Details
| Original language | English |
|---|---|
| Title of host publication | Conference Record of the 58th Asilomar Conference on Signals, Systems and Computers, ACSSC 2024 |
| Editors | Michael B. Matthews |
| Pages | 1427-1432 |
| Number of pages | 6 |
| ISBN (electronic) | 9798350354058 |
| Publication status | Published - 27 Oct 2024 |
| Event | 58th Asilomar Conference on Signals, Systems and Computers, ACSSC 2024 - Pacific Grove, United States Duration: 27 Oct 2024 → 30 Oct 2024 |
Publication series
| Name | Conference Record - Asilomar Conference on Signals, Systems and Computers |
|---|---|
| ISSN (Print) | 1058-6393 |
Abstract
Cochlear implants (CIs) are surgically implanted hearing devices, which allow to restore a sense of hearing in people suffering from profound hearing loss. Wireless streaming of audio from external devices to CI signal processors has become common place. Specialized compression based on the stimulation patterns of a CI by deep recurrent autoencoders can decrease the power consumption in such a wireless streaming application through bit-rate reduction at zero latency. While previous research achieved considerable bit-rate reductions, model sizes were ignored, which can be of crucial importance in hearing-aids due to their limited computational resources. This work investigates maximizing objective speech intelligibility of the coded stimulation patterns of deep recurrent autoencoders while minimizing model size. For this purpose, a pruning-aware loss is proposed, which captures the impact of pruning during training. This training with a pruning-aware loss is compared to conventional magnitude-informed pruning and is found to yield considerable improvements in objective intelligibility, especially at higher pruning rates. After fine-tuning, little to no degradation of objective intelligibility is observed up to a pruning rate of about 55 %. The proposed pruning-aware loss yields substantial gains in objective speech intelligibility scores after pruning compared to the magnitude-informed baseline for pruning rates above 45 %.
Keywords
- autoencoders, cochlear implants, pruning, pruning-aware loss, stoi, vstoi, wireless transmission
ASJC Scopus subject areas
- Computer Science(all)
- Signal Processing
- Computer Science(all)
- Computer Networks and Communications
Sustainable Development Goals
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Conference Record of the 58th Asilomar Conference on Signals, Systems and Computers, ACSSC 2024. ed. / Michael B. Matthews. 2024. p. 1427-1432 (Conference Record - Asilomar Conference on Signals, Systems and Computers).
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Pruning-Aware Loss Functions for STOI-Optimized Pruned Recurrent Autoencoders for the Compression of the Stimulation Patterns of Cochlear Implants at Zero Delay
AU - Hinrichs, Reemt
AU - Ostermann, Jörn
N1 - Publisher Copyright: © 2024 IEEE.
PY - 2024/10/27
Y1 - 2024/10/27
N2 - Cochlear implants (CIs) are surgically implanted hearing devices, which allow to restore a sense of hearing in people suffering from profound hearing loss. Wireless streaming of audio from external devices to CI signal processors has become common place. Specialized compression based on the stimulation patterns of a CI by deep recurrent autoencoders can decrease the power consumption in such a wireless streaming application through bit-rate reduction at zero latency. While previous research achieved considerable bit-rate reductions, model sizes were ignored, which can be of crucial importance in hearing-aids due to their limited computational resources. This work investigates maximizing objective speech intelligibility of the coded stimulation patterns of deep recurrent autoencoders while minimizing model size. For this purpose, a pruning-aware loss is proposed, which captures the impact of pruning during training. This training with a pruning-aware loss is compared to conventional magnitude-informed pruning and is found to yield considerable improvements in objective intelligibility, especially at higher pruning rates. After fine-tuning, little to no degradation of objective intelligibility is observed up to a pruning rate of about 55 %. The proposed pruning-aware loss yields substantial gains in objective speech intelligibility scores after pruning compared to the magnitude-informed baseline for pruning rates above 45 %.
AB - Cochlear implants (CIs) are surgically implanted hearing devices, which allow to restore a sense of hearing in people suffering from profound hearing loss. Wireless streaming of audio from external devices to CI signal processors has become common place. Specialized compression based on the stimulation patterns of a CI by deep recurrent autoencoders can decrease the power consumption in such a wireless streaming application through bit-rate reduction at zero latency. While previous research achieved considerable bit-rate reductions, model sizes were ignored, which can be of crucial importance in hearing-aids due to their limited computational resources. This work investigates maximizing objective speech intelligibility of the coded stimulation patterns of deep recurrent autoencoders while minimizing model size. For this purpose, a pruning-aware loss is proposed, which captures the impact of pruning during training. This training with a pruning-aware loss is compared to conventional magnitude-informed pruning and is found to yield considerable improvements in objective intelligibility, especially at higher pruning rates. After fine-tuning, little to no degradation of objective intelligibility is observed up to a pruning rate of about 55 %. The proposed pruning-aware loss yields substantial gains in objective speech intelligibility scores after pruning compared to the magnitude-informed baseline for pruning rates above 45 %.
KW - autoencoders
KW - cochlear implants
KW - pruning
KW - pruning-aware loss
KW - stoi
KW - vstoi
KW - wireless transmission
UR - http://www.scopus.com/inward/record.url?scp=105002677761&partnerID=8YFLogxK
U2 - 10.1109/IEEECONF60004.2024.10943066
DO - 10.1109/IEEECONF60004.2024.10943066
M3 - Conference contribution
AN - SCOPUS:105002677761
T3 - Conference Record - Asilomar Conference on Signals, Systems and Computers
SP - 1427
EP - 1432
BT - Conference Record of the 58th Asilomar Conference on Signals, Systems and Computers, ACSSC 2024
A2 - Matthews, Michael B.
T2 - 58th Asilomar Conference on Signals, Systems and Computers, ACSSC 2024
Y2 - 27 October 2024 through 30 October 2024
ER -